Chiron

Chiron performs end-to-end basecalling of raw electrical signals from Oxford Nanopore Technologies (ONT) sequencers to produce DNA sequences using deep learning.

Key Features:

• ONT raw-signal basecalling: Performs basecalling on raw electrical signals generated by Oxford Nanopore Technologies (ONT) nanopore sequencers.
• End-to-end basecalling: Directly translates raw signals into nucleotide sequences without an intermediary segmentation step.
• Deep learning model: Uses a deep learning model trained on 4,000 reads to predict nucleotide sequences.
• Cross-species generalization: Demonstrates generalization to species not present in the training set.
• High throughput: Processes over 2,000 bases per second on desktop GPUs.
• High accuracy: Demonstrates high basecalling accuracy from noisy nanopore signals.

Scientific Applications:

• Clinical diagnostics: Enables rapid DNA sequence generation useful for clinical diagnostic workflows.
• Evolutionary biology: Supports sequencing tasks in evolutionary biology studies.
• Genomics research: Facilitates rapid sequencing workflows in diverse genomics research applications requiring fast basecalling.

Methodology:

Chiron employs a deep learning model that performs end-to-end basecalling directly from raw nanopore electrical signals, was trained on 4,000 reads, and runs inference on desktop GPUs at over 2,000 bases per second.